10 Abstract Actinobacteria are important bacterial group participating in various ecosystem processes particularly in the decomposition of complex organic compounds. Their abilities enable them to surviving in harsh conditions of oligotrophic habitats like lakes, deserts, cave walls or recalcitrant and resistant litter in soil, where Actinobacteria often dominate. Although certain biotic and abiotic factors were recognized to modulate Actinobacteria incidence in such habitats, the influence of anthropogenic pressure on their communities is scarcely known. The main objective of this thesis is therefore to determine differences of Actinobacteria communities under the direct (the human visitors changing microenvironment of caves, part 1) and indirect (climate change factors like altered precipitation or plant litter quality, part 2) anthropogenic influence in two habitats, plant litter in soil and cave walls, where Actinobacteria play important roles and dominate. In a first part of the thesis we monitored Actinobacteria communities in French limestone caves walls differently affected by humans (pristine versus anthropized caves). For identification of important species like potential pathogens or pigments producing Actinobacteria using amplicon sequencing of environmental DNA (Illumina MiSeq), we firstly used...

L'évaluation écologique des cours d'eau ne prend pas en compte les différents processus qui y sont réalisés, faute d'indicateur disponibles. L'objectif de cette thèse a été de documenter l'utilisation de la décomposition des litières végétales d'origine terrestre comme indicateur de l'intégrité fonctionnelle des rivières de plaine et de piedmont suivies par l'Agence de l'Eau Adour-Garonne. Partie 1. Sur une douzaine de sites suivis pour la pollution par les pesticides, le taux de décomposition est divisé par deux entre le site le moins et le plus contaminé, entrainant le passage d'une décomposition co-assurée par les invertébrés et les microorganismes dans les sites les moins contaminés à un processus réalisé seulement par les microorganismes dans les sites les plus contaminés, compromettant le processus dans les rivières les plus contaminées. Cette altération est observée à des concentrations en pesticides pourtant inférieures aux seuils fixés par l'Union Européenne. Partie 2. Sur 58 sites qui maximisent l'amplitude des classes de qualité écologiques, l'absence de corrélation observée entre les indices biologiques basés sur les invertébrés (IBGN, I2M2) et les diatomées (IBD) et le processus de décomposition suggère que les perturbations, selon leur nature et/ou leur intensité, altèrent différentiellement la composition des communautés et les activité de transformation de la matière organique. Partie 3. Une grille d'évaluation de la qualité fonctionnelle des rivières croisant deux variables, ktotal et le ratio kinvertébré/kmicrobien est proposée à partir de 84 sites de suivi et de sites de référence tirés de la bibliographie. Cette grille montre qu'à l'échelle régionale, le processus de décomposition est plus affecté dans sa composante liée à l'activité des invertébrés qu'à celle liée aux microorganismes. En suivant les critères d'évaluation de la Directive Cadre sur l'Eau, cette grille d'évaluation met en évidence que seulement 28 % des sites étudiés atteindrait un bon état fonctionnel. Cette thèse a permit de documenter la réflexion menée sur le développement d'indicateurs fonctionnels et plaide pour l'inclusion de la décomposition des litières végétales dans l'évaluation évaluation écologique par une approche croisée, telle que formalisée par un graphique structure x fonction à double entrée. / The aim of this thesis was to document, at regional scale (SW France), the application of litter decomposition and associated metrics as a proxy of functional integrity of stream ecosystems. The Introduction exposed the underlying arguments, mainly relying on (1) the importance of this ecosystem process, which provides most carbon sources to the aquatic food web, and (2) the need to complement the structural metrics routinely used in biomonitoring schemes by a functional component. Chapter I dealt with the capacity, by using litter decomposition, to discriminate the effects of pesticides from other stressors in 12 agricultural streams selected along a contamination gradient. The decomposition rate was twice as low in the most contaminated site compared with the less contaminated one, and the relationship was monotonous along the gradient. Only the activity of invertebrate decomposers was affected, translating into the switch from a decomposition co-dominated by invertebrates and microorganisms in the less contaminated sites to a process only achieved by microorganisms in the most contaminated ones. Taxonomic richness and density of invertebrate decomposers together with a specific indicator based on invertebrate traits followed the same trends as the invertebrate-driven decomposition rate, kinvertebrate, while fungal biomass, sporulation rate and taxonomic richness remained unaffected. Consequently, the presence of pesticides compromised leaf breakdown, as microbial decomposers did not compensate for the invertebrate decomposers decline. This occurred while pesticides concentrations even in the most contaminated stream were under the European Union's Uniform Principles thresholds for targeted species. This study showed that litter breakdown, particularly the ratio of total to microbial-driven breakdown rate, is a pertinent proxy to assess the functional integrity of pesticide-contaminated streams. Chapter II compared the biological evaluation, as routinely undertaken in monitoring schemes from structural indicators based on invertebrates (IBGN and I2M2) and diatoms (BDI), with the functional indication based on litter decomposition. No correlation was found between structural and functional indicators on 58 streams, whereas the 3 structural indicators were strongly correlated each other. Redundancy analysis showed a negative relationship between ktotal or kinvertebrate and the proportion of cultures in the riparian corridor, while kmicrobial responded to the same environmental parameters (substratum coarseness, temperature) as structural indicators. The lack of correlation between invertebrate-based structural metrics and kinvertebrate suggested that disturbances, depending on their nature and/or intensity, differentially impair community composition and community activity in organic matter processing. In the view of proposing functional metrics, Chapter III concerned the distribution patterns of decomposition parameters on our study sites together with reference sites from the literature. Strong links were detected between decomposers community, decomposition rate and environmental settings, as exemplified by positive correlations between ktotal or kinvertebrate and detritivores abundance, kmicrobial and fungal biomass or sporulation rate, and the ratios ktotal/kmicrobial or kinvertebrate/kmicrobial and the abundance of decomposers. We propose a rating grid of stream ecosystem functional quality crossing two entries, ktotal and kinvertebrate/kmicrobial. At the regional scale, this grid shows that litter decomposition is more affected in its invertebrate component than in the microbial one. The Discussion exposes the main results in context and opens to perspectives, particularly about the question of reference sites. In conclusion, our results strongly argue for a crossed evaluation of ecological quality of running waters, as formalized by a chart with structure × fonction double entry.